Cocaine use disorder (CUD) is a significant global health problem, affecting millions worldwide. Characterized by compulsive cocaine use, intense craving, and high relapse rates, CUD lacks effective pharmacotherapy, with treatment primarily focused on symptom management. Neurobiological alterations in the brain are believed to underlie CUD's clinical manifestations, supported by neuroimaging studies showing structural and functional changes, particularly in the frontal cortex (involved in inhibitory control) and striatal regions (involved in reward processing). Epigenetic and gene expression changes are hypothesized as molecular mechanisms driving these brain alterations. While research in rodent models has identified differentially methylated regions (DMRs) and differentially expressed genes (DEGs) in brain regions like the prefrontal cortex (PFC) and nucleus accumbens (NAC), human studies are limited. Existing human studies, using postmortem brain tissue, have shown some evidence of altered DNA methylation and gene expression in CUD, but these have typically focused on either DNA methylation or gene expression in isolation, or used relatively small sample sizes. The prefrontal cortex, particularly Brodmann Area 9 (BA9), is of interest due to its role in executive control and its involvement in the preoccupation/anticipation stage of addiction. Furthermore, the relationship between epigenetic modifications, such as DNA methylation (DNAm), and gene expression in the same brain samples remains largely unexplored in the context of CUD. Beyond DNA methylation and gene expression, alternative splicing (AS) may also contribute to neurobiological changes in CUD, as has been demonstrated in other substance use disorders (SUDs). Alterations in AS have been reported in alcohol use disorder (AUD) and opioid use disorder (OUD), indicating that this mechanism may be a common feature of SUDs. However, comprehensive analyses of AS in human CUD have been lacking. This study aimed to bridge this gap by using a multi-omics approach to investigate the molecular underpinnings of CUD in the human prefrontal cortex.

Several studies have investigated the epigenetic and transcriptional changes in the brains of individuals with CUD, mostly using rodent models. These studies have shown that cocaine exposure can lead to changes in DNA methylation and gene expression in various brain regions, including the PFC and NAC. Specifically, studies using reduced representation bisulfite sequencing (RRBS) have identified several DMRs in the NAC and caudate nucleus of individuals with CUD. Other studies have reported transcriptome-wide gene expression changes in the same brain regions, highlighting alterations in synaptic transmembrane transporter genes and immune processes. One of the largest studies investigating transcriptomic changes in the PFC identified a significant number of DEGs in Brodmann Area 46, with enriched pathways related to GTPase signaling and neurotransmitter secretion. Prior research from the authors also revealed CUD-associated DMRs in BA9, with co-methylation networks enriched for synaptic signaling processes. However, the integration of epigenomic and transcriptomic data from the same individuals, along with an assessment of alternative splicing, has been largely lacking, limiting a comprehensive understanding of the molecular mechanisms involved in CUD.

This study utilized postmortem brain tissue from the Douglas Bell Canada Brain Bank (DBCBB). The cohort comprised 42 male individuals of European American descent diagnosed with cocaine use disorder (CUD) based on DSM-5 criteria. Control individuals were matched for age, sex, and ancestry. Exclusion criteria included severe neurodevelopmental or psychiatric disorders (excluding depressive disorders) and additional SUDs (except AUD). DNA was extracted from Brodmann Area 9 (BA9) tissue samples, and DNA methylation was profiled using the Illumina Methylation EPIC BeadChip. RNA was extracted from a subset (N=25) of BA9 samples, with RNA sequencing (RNA-seq) performed after rRNA depletion. Data analysis included: * **DNA methylation analysis:** Preprocessing using an in-house quality control (QC) pipeline, neuronal cell fraction estimation, quantile normalization, and EWAS using a linear regression model adjusted for covariates (age, PMI, pH, neuronal cell fraction, comorbidities). Downstream analysis included DMR identification, gene ontology (GO) enrichment analysis, and weighted gene co-expression network analysis (WGCNA). * **Gene expression analysis:** Quality control using FastQC, read mapping to the GRCh38 genome using STAR, quantification using featureCounts, and differential expression (DE) testing using DESeq2. The DE analysis was adjusted for covariates (age, PMI, pH, RNA integrity number (RIN)). Fold-change and p-value cutoffs were used to define DEGs. Cell type deconvolution was performed using CIBERSORT. Functional enrichment analysis was conducted using GSEA, and WGCNA was used to identify CUD-associated co-expression modules. * **Alternative splicing analysis:** Performed using LeafCutter, an annotation-free approach for quantifying differential splicing. Data was aligned to the GRCh38 genome using STAR, and differential intron excision analysis was performed, adjusting for covariates (age, PMI, pH, RIN). * **Replication analysis:** CUD-associated DEGs were investigated in two independent datasets: BA9 bulk RNA-seq data from the National PTSD Brain Bank (NPBB) and neuronal-specific RNA-seq data from Brodmann Area 46 (BA46) from the University of Miami Brain Bank (MBB). * **Drug repositioning analysis:** Connectivity Map (CMap) was used to identify drugs that could reverse the CUD expression profile. The top upregulated and downregulated genes were used as input. * **Integrative analysis:** Data from DNA methylation, gene expression, and alternative splicing analyses were integrated using Spark and MOFA. An integrative functional GO enrichment analysis was also performed to identify converging pathways across all omics layers.

The transcriptome-wide analysis identified 1057 DEGs (p<0.05), with ZFAND2A significantly upregulated (q<0.05) in CUD. Differential alternative splicing analysis revealed 108 significantly differentially spliced intron clusters (FDR<0.05) across 98 genes, enriched for pathways involved in cell junction formation and neuron projection extension. Replication analysis in two independent cohorts showed convergent results for HSPA6 and FKBP4 as upregulated genes. Two genes, ZBTB4 and INPP5E, displayed consistent alterations across DNA methylation, gene expression, and alternative splicing analyses. Pathway analyses revealed convergence in pathways related to synaptic signaling, neuron morphogenesis, and fatty acid metabolism across the various omics layers. Drug repositioning analysis implicated glucocorticoid receptor-targeting drugs as potential candidates for reversing the CUD expression profile. Multi-omics factor analysis identified a factor significantly correlated with CUD, which was enriched for synaptic signaling, cell junction organization, and neurogenesis pathways.

This multi-omics study provides novel insights into the molecular mechanisms underlying CUD in the human prefrontal cortex. The identification of consistently altered genes (ZBTB4 and INPP5E) across different molecular layers highlights the complexity of CUD pathophysiology. The convergence of findings in synaptic signaling pathways is consistent with previous research emphasizing the role of neurotransmission in CUD. The observed alterations in neuronal morphology and fatty acid metabolism suggest a more comprehensive view of the disease process. The identification of glucocorticoid receptor-targeting drugs as potential therapeutic agents provides a promising avenue for future research. The finding of alterations in alternative splicing adds another layer of complexity to the molecular changes associated with CUD, which could potentially contribute to the neuroplastic changes associated with addiction.

This study presents a comprehensive multi-omics analysis of CUD in the human prefrontal cortex, revealing novel molecular signatures. The consistent alterations in genes like ZBTB4 and INPP5E, along with the convergence of pathways related to synaptic signaling and neuronal morphology, provide important insights into CUD pathophysiology. The identification of glucocorticoid receptor-targeting drugs warrants further investigation as potential therapeutic agents. Future studies should address the limitations of this study, especially by including larger and more diverse cohorts to validate the findings and investigate sex and ancestry-specific differences.

This study's cross-sectional design limits the ability to identify dynamic changes in DNA methylation and gene expression throughout the disease course. The relatively small sample size may restrict the generalizability of the findings, although replication analysis in independent cohorts provides some support. The study population's homogeneity (only males of European American descent) limits the ability to explore sex-specific and ancestry-related molecular signatures. While some relevant clinical variables (e.g., cocaine at death and cause of death) were not included in analyses due to statistical power limitations, future research with larger sample sizes is needed to assess their influence. Finally, the study did not examine protein expression levels, which would provide a more direct link between mRNA expression and functional consequences.